- Italy’s Consiglio Nazionale delle Ricerche (CNR) updated its policy templates in 2025 through its Research Ethics and Integrity Committee, covering incidental findings management.
- CNR requires ethical clearance for all research projects before approval or funding, including national, European, and international calls, making ethics review a structural gate.
- CNR has established a systematic institutional framework for ethical AI oversight, featuring mandatory clearance, structured training, and ongoing legal consultation.
Italy’s Consiglio Nazionale delle Ricerche has built one of Europe’s more systematic institutional frameworks for ethical AI oversight, one that operates not through post-hoc review but through mandatory clearance, structured training and ongoing legal consultation. A 2025 update to its policy templates for managing incidental findings, issued by the CNR’s Research Ethics and Integrity Committee, is the latest iteration of an approach the institution has been developing for years. Whether that framework is as effective in practice as it is on paper is a question the available public material does not fully answer.
Mandatory Ethical Clearance for AI Projects
The practical centre of CNR’s approach is its ethical clearance requirement. Researchers at CNR, and external collaborators working with the institution, must submit projects to the Research Ethics and Integrity Committee before receiving approval or funding. The process applies to national, European and international funding calls, meaning the ethics gate is structural rather than advisory. The stated aim is to identify potential issues at the design stage rather than after problems emerge, what CNR terms an “ethics-by-design” approach. How consistently that standard is applied across CNR’s many institutes is harder to assess from public documentation.
Developing Comprehensive Research Integrity Guidelines
The most substantial is its “Guidelines for Research Integrity.” A separate “Code of Ethics and Deontology for Scientific Activity in the field of Cultural Heritage” addresses domain-specific concerns.
Proactive Prevention of Research Misconduct
The Committee also runs a prevention and verification programme aimed at research misconduct. This includes ethics consulting for cases involving CNR staff or external personnel, and preventive technical advice to researchers ahead of article or project submissions, specifically to reduce unintentional fabrication, falsification or plagiarism of data, images or text. The emphasis on unintentional misconduct is notable: much of the framing addresses researchers making honest errors rather than deliberate fraud, which reflects the practical reality of integrity issues in large research institutions. Whether the consulting capacity scales adequately across all CNR departments is not addressed in the available public material.
Continuous Training and Dissemination
CNR runs periodic training activities for researchers across its local offices, supplemented by conferences, seminars and educational materials. The programme is designed to track the evolving ethics landscape, including developments in AI-relevant research areas. For an institution operating across disciplines as varied as materials science, social research and computer science, maintaining a training programme that genuinely addresses AI-specific ethics questions, rather than generic research integrity principles, is a non-trivial challenge. The public record describes the programme’s existence; its depth and uptake are less visible. For context on how web-connected AI systems are creating new governance challenges that training programmes need to address, see our coverage of OWASP excessive agency risks in AI agents.
Specialised Ethical-Legal Consulting and Monitoring
CNR’s “Ethics Mentoring” service provides ethical and legal supervision of active research projects, typically at the request of funding agencies. The Committee also advises CNR departments and institutes on complex ethical-legal questions as they arise during research. This layer of ongoing consultation addresses a genuine gap in many institutional frameworks, where ethics review stops at project approval rather than continuing through a project’s lifecycle. How frequently the service is used, and whether it reaches the institution’s AI research specifically, is not detailed in publicly available documentation. The broader question, whether CNR’s framework is a model worth examining for other European research institutions navigating the EU AI Act’s compliance requirementsis one the institution’s 2025 updates may be positioning it to answer. For more coverage of AI policy and regulation, visit our AI Policy & Regulation section.
